Pages: 1 [2]   Go Down
Author Topic: Another decoupling question...?  (Read 1258 times)
0 Members and 1 Guest are viewing this topic.
0
Offline Offline
Edison Member
*
Karma: 7
Posts: 1221
Arduino rocks
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

so to use a ceramic cap is not enough.. or not 'correctly' decoupling the power lines...correct?

it 'must' use a electrolytic cap.. (for the polarity)..


thanks..

(guess my board(s) are on the verge of maybe just NOT/STOP working one day?)  (yikes!)



Logged


Manchester (England England)
Online Online
Brattain Member
*****
Karma: 509
Posts: 31480
Solder is electric glue
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

The ratio of large to small caps is not large. One large cap for every ten or so small ones.
Remember caps are only one aspect of decoupling there is seriese resistors and inductors to consider as well.
An under decoupled system might work but be more prone to disruption by interference. It is only when you start testing the system in the presence of interference can you tell how robust it is going to be.
Getting enough decoupling in the right place accounts for over 85% of all problems in professional development of cutting edge products and warps responsible for more board revisions than anything else.
Logged

Offline Offline
Edison Member
*
Karma: 116
Posts: 2205
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Quote
so to use a ceramic cap is not enough

There are two types of decoupling: bulk decoupling and (device) decoupling. We are talking about the later. For that, you want to have a small but fast capacitor (ceramic or polyethylene). A typical value is 104 monolithic. However, I typically use a 4.7uf electrolytic (for ease).

Logged

the land of sun+snow
Offline Offline
Faraday Member
**
Karma: 149
Posts: 2790
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Quote
elec caps are SO big!.....

There are smt electrolytics that are not very large, even for good-sized values like 10...22 uF.

Quote
there is no Electrolytic caps even being used  (guess that is in incorrect).. I didnt see any +/- in the schematics..where others 'did' have it

People do everything, but as indicated, the best practice is to use both. OTOH, you also need to
check what the datasheet says [as others said]. smt devices may have special directions, such as
cap values within certain ranges.

Quote
I have had numerous systems running without decoupling caps.

Ditto to the first sentence of the previous answer.

Quote
so to use a ceramic cap is not enough.. or not 'correctly' decoupling the power lines...correct?

it 'must' use a electrolytic cap.. (for the polarity)..

I believe the term 'decoupling' usually refers to the bypass caps [small-valued ceramics]. Also,
tantalum caps are a replacement for electrolytics, and have much better high-frequency
characteristics, but generally cost 20X as much more more.
Logged

Anaheim CA.
Offline Offline
Faraday Member
**
Karma: 44
Posts: 2810
...
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

In the upper right corner of the board is a small yellow block with a brown band on one end... That Sure looks like a tantalum capacitor or at least it looks Exactly the same as the tantalums I used in place of smd electrolytics. for coupling and bypass. Typically I would start with a 220 - 470uF elect. cap on the battery and a 100 uF elect. cap on the output of the regulator MAX683??? I really don't remember but it was an electrolytic and I just wanted 'something' there. I should stop and note that Electrolytics should be about twice as big as calculated because over any appreciable time they will always decrease in value and decrease value with temperature.  MLM Ceramic capacitors were relatively new and expensive in 2008 when I retired and I used tantalums whenever I needed a stable and good quality capacitor. I would use at least one 10uF tant or possibly a 47uF tant. Cap on each design along with a 100 nF per IC. From personal experience it pays to be as conservative as possible. There is another thought too... add as many as you have reasonable room for... You don't have to populate the positions of unnecessary parts... and should the part 'somehow' be required... You've a place for it.
Finally the parts are totally invisible, inexpensive, insurance and will not limit the rise time of the regulator output voltage... unless the current limit of the IC is exceeded.

Bob
Logged

--> WA7EMS <--
“The solution of every problem is another problem.” -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

UK
Offline Offline
Faraday Member
**
Karma: 92
Posts: 3969
Where is your SSCCE?!?!
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Yep - I have some of the exact same tantalums here.  That looks rather like a 47µF one to me, but it's hard to make out the markings on that photo.

And yes, pads are free, but components cost.  Adding pads where they might be wanted in the future is always good.  It doesn't cost any more, and could save a whole bundle on a board re-design.


* tants.jpg (30.64 KB, 480x324 - viewed 3 times.)
« Last Edit: November 07, 2012, 05:16:46 am by majenko » Logged

Why not visit my eBay shop? http://stores.ebay.co.uk/Majenko-Technologies
Replacement for the Arduino IDE: UECIDE - Proper serial terminal, graphing facilities, plugins, overhauled internals.
Java isn't bad in itself, but it has enabled morons to write programs.

0
Offline Offline
Edison Member
*
Karma: 7
Posts: 1221
Arduino rocks
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

sorry,.. to be clear.. there isnt any (elect/tant caps on the power rails...

that one tant cap is used for the blocking ob the audio/speaker output line..
Logged


Valencia, Spain
Offline Offline
Faraday Member
**
Karma: 118
Posts: 4554
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

It is rare that there is too much decoupling. The only case I have ever encounters is that some DC to DC converters will not start with too much of a capacitave load, and the odd case of low dropout regulators, most are not that fussy.

If you put too much in isn't it a bit like shorting out the power supply at startup?

Logged

No, I don't answer questions sent in private messages...

the land of sun+snow
Offline Offline
Faraday Member
**
Karma: 149
Posts: 2790
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Quote
It is rare that there is too much decoupling. The only case I have ever encounters is that some DC to DC converters will not start with too much of a capacitave load, and the odd case of low dropout regulators, most are not that fussy.

If you put too much in isn't it a bit like shorting out the power supply at startup?

It depends upon how you define decoupling, eg small ceramic caps or large electrolytics. The term has
been used both ways on this thread. The following people use decoupling to refer to the former, small
ceramics.

http://www.learnemc.com/tutorials/Decoupling/decoupling01.html
http://www.hottconsultants.com/techtips/decoupling.html

A lot of small value caps will likely have little effect on startup. OTOH, if the values of the electrolytics
are too high, then the power supply may be slow in starting up, and possibly affect bootup of the
microcontroller, but I wouldn't think this would be a problem in general, unless the values are really
high, say over 100 or 500 uF, depending on the circuit.

The only reference I could find to this in the 328P d/s was the following, and they're not referring to
boot-up, although by heavily filtered they certainly mean the electrolytics.

"In heavily filtered power supplies, VCC is likely to rise or fall slowly on power-up/down. This causes
the device for some period of time to run at a voltage lower than specified as minimum for the clock
frequency used."

Logged

Anaheim CA.
Offline Offline
Faraday Member
**
Karma: 44
Posts: 2810
...
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Well Yeah that's electronics 101... But the only way to have a "Slow" power supply is to design one that way, by using 1. Low current linear regulators (I've used 2200uF on the output side of a 7805 and the BOD working on the processor, Very successfully). 2. By adding large (>10 ohm series "Filter" R's in the Vcc line, while a common method for isolating low power loads, should never be used for the main Vcc unless the load can tolerate being powered up "later".  and 3. using a primary power source that can't fully supply the required load current. In the past I've successfully used 470uF caps with 100 mA regulators both 78L05's as well as LM2931's to power PIC processors in low speed low power applications with great success with the BOD enabled. The Basic Layout I used was a 100 to 470uF electrolytic right where the battery connected to the PCB. This part isn't so much filtering or 'decoupling' as it is compensation for a low battery charge condition, because the batteries internal resistance increases when the battery is nearly discharged and the addition of this part enables the use of a little more of the power in the battery. After the linear regulator I would use between 47 to 220 uF (depending on what the board had to do) a 10 to 22uF tant cap somewhere in the "middle" of things and a 100nF cap for each IC right at the IC. What works in one application usually won't work in production and skimping on bypassing because you don't fully understand the topic is usually a waste of time because the circuit might work well in the test configuration usually doesn't when everything is interconnected. You will have greater success if you remember that...
Murphy was an optimist..

Bob
Logged

--> WA7EMS <--
“The solution of every problem is another problem.” -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

Pages: 1 [2]   Go Up
Jump to: